Re: Some good book or guide to understand statistics in deep?

Thanks Anjo, my point is I want for example to evaluate the new row migration in 10g compared with moving to another tablespace, solution to reduce table high watermark.
( there are several situations where youcan't use sql*plus statistics)

Then I use Tom Kyte runstats, the problem, is I saw a group of statistics I have no idea what are about. I go to the documentation and read the descripction, and I understand what they show, but what I don't understand is 1) if this is a problem or is harmless
2) how can I know if this is really a problem Honestly I don't have time to digg to every one statistic and test and understand.

For example the hard parse statistics, I know this can be a problem, if this is more than the distinct sqls running in this period, approximately I can estimate.

I'm looking something like this, or better. parse count (hard)
This statistic shows hard parses (a high cpu consuming process, low hard disk consuming, medium latch consuming). The value should be approximately the same number than distinct sql run by first time.

Something that givesme confidence to kwno if I'm interpreting correctly what I see.

I think there should be something like this in some place. I want cpu, disk and latch consumption for every statistics to evaluate if this is important or not

:)

--
http://www.freelists.org/webpage/oracle-l